JP2005352878A - Document retrieval system, retrieval server and retrieval client - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide the summary of a retrieval result in an associative retrieval system based on a plurality of view points. <P>SOLUTION: One document database is indexed in a plurality of manners, whereby a retrieval result can be overview-displayed from a plurality of view points. Respective documents in indexed document databases 403, 503 and 603 are managed with a common identifier, whereby the summary of a document group obtained as the retrieval result can be formed by use of the respective indexes. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a document search system, and more particularly to an associative search system that displays search results from a plurality of viewpoints.

  With the spread of computers and the Internet, the digitization of document information is progressing rapidly. As the available information increases, it becomes an important issue to search for necessary information. In addition, there is an increasing demand for examining the relationship between document groups among a plurality of document databases. For example, there are many requests to search for articles of encyclopedia related to newspaper articles of interest.

  In keyword search that is currently in practical use, it is possible to search by switching between multiple document databases. However, for a document group included in a document database, a related document group is assigned to the same document database, or It is impossible to search from another document database (search method called document associative search).

  If it is limited to the same document database, it is possible to realize a document associative search using a document group as a search input by calculating the degree of association between documents in advance. However, for a plurality of document databases, the degree of association between documents to be calculated in advance increases as the number of document databases increases.

  On the other hand, Japanese Patent Laid-Open No. 2000-155758, “Document Search Method and Document Search Service Targeting Multiple Document Databases”, describes an arbitrary document group in a document database designated by a user for the document. A method for efficiently retrieving a document group related to a group from an arbitrary document database is disclosed. This method realizes a high-speed document associative search by using only characteristic words in the search input inputted as a document group. With this method, the user can perform highly accurate and efficient document search by checking the relationship between document groups while switching between a plurality of different types of document databases. In this method, characteristic words appearing in the document group obtained as a search result are extracted and presented to the user as an overview (summary) of the search result. It also provides a support means when judging.

JP 2000-155758 A

  In document retrieval based on words, the documents are indexed by using words appearing in the document to realize document retrieval. The same is true for the method disclosed in Japanese Patent Laid-Open No. 2000-155758, and in order to extract characteristic words from a document, a statistical measure (such as the tf * idf method) is used for the words included in the document. The degree of importance is calculated by using (a), and words are extracted in descending order of importance. In general, indexing is performed on a single document database in one way. However, technical terms (such as disease names, gene names, and protein names in the biomedical field) and fact information (for example, protein-protein interactions in the biomedical field) are buried in the general word distribution. It becomes difficult to be extracted as a feature word. In addition, when indexing is complete, the search result overview is limited to one viewpoint, so if that viewpoint does not match the user's search request or interest Cannot be an appropriate overview display.

  An object of the present invention is to provide a document search system that displays an overview of search results from a plurality of viewpoints that match a user's interest.

  In order to solve the above-described problems, in the present invention, a search result can be displayed in an overview from a plurality of viewpoints by indexing a single document database in a plurality of ways.

  For example, one word database is indexed with ordinary words, indexed with technical terms, and indexed with fact information. In order to correspond to each indexed document database, each document is managed by a common identifier so that an outline from a document can be created using each index.

  A document search system of the present invention includes a search client including an input unit for inputting a search request, a search result display unit for displaying a searched document group, and a general display unit for displaying an outline of the searched document group, and an index A document database that stores a plurality of attached documents, a search unit that searches the document database for documents that are highly relevant to the received search request, and an outline for a given document group using the index And a search server including a schematic creation unit to be created, and includes a plurality of different types of indexes as indexes.

  On the summary display section of the search client, a plurality of types of summary are displayed separately for each different viewpoint. The search result display unit has a document selection unit for selecting a document to be a key for the next search from the displayed document group, and the schematic display unit displays the next of the displayed schematic elements. It has a rough selection unit for selecting an element that is a key for search.

  By viewing the overview display from a plurality of viewpoints with respect to the document set obtained as a search result, the user can more appropriately grasp the nature of the search result. Further, since the relationship between the viewpoints can be regarded as a search target document, the search result can be analyzed in more detail.

  Embodiments of the present invention will be described below with reference to the drawings.

  FIG. 1 is a schematic diagram showing a configuration example of a system for realizing the present invention. This system includes a search client 20 in which a user inputs a search request and displays a search result, search servers 40, 50, 60 for searching a document database, search client 20 and search servers 40, 50, The associative search server 30 that mediates 60 is connected by the communication network 10. In the example shown in the figure, three search servers are connected to the communication network as search servers for searching the document database, but the number of search servers connected to the communication network may be arbitrary. The number of search clients is also arbitrary.

  The search means 402, 502, 602 provided in the search servers 40, 50, 60 search the document database 403, 503, 603 for a document group having a high degree of relevance in response to the search request sent from the associative search server. The search result is returned to the associative search server 30 with the weight of relevance. The search means here can be realized by, for example, a known keyword search method.

  In the keyword search method, in order to increase the efficiency of the search process, the document contained in the document database is divided into words (morphological analysis is performed for Japanese documents, and stemming processing is performed for English documents) Which word is included in which document is created as an index. When the search is executed, the created index is read into the main memory, so that the search process can be executed at high speed. In FIG. 1, indexes 404, 504, and 604 are created for document databases 403, 503, and 603 of search servers 40, 50, and 60, respectively, and are used for search processing.

  The outline creation means 401, 501, and 601 included in the search servers 40, 50, and 60 create outlines of document groups searched from the document databases 403, 503, and 603. Here, an outline is a word set that well represents the contents of a document group. As the outline creation means, an existing method such as JP 2000-155758 A can be used. The above index is also used when creating the outline. That is, it is checked by referring to the index which word is included in a certain document.

  As an example, first, the frequency of words included in all the documents in the document group to be outlined is totaled. In general, the more frequently a word appears in a document group, the higher the degree of representation of the document group. Therefore, the higher the frequency of appearance in the document group, the easier it is to be included in the outline. However, general words that frequently appear in any document, such as “do”, are not suitable as approximate words. Therefore, generally, an approximate word is selected in consideration of the appearance frequency in the document database to which the document group belongs. That is, a word having a higher appearance frequency in a specified document group and a lower total appearance frequency in the entire document database is a characteristic word in the sense that it appears only in the document group. It is suitable as an outline to characterize. Specifically, for each word in the document group, the weight of the word is calculated by an appropriate function that inputs the appearance frequency in the document group and the appearance frequency in the document database. Adopt the word you have as an outline.

  The search client 20 includes search request input means 201, search result display means 202, and summary display means 203.

  FIG. 2 is a diagram illustrating an example of an initial screen in the search client. A user inputs a search request in the search request input area 2011 and clicks a search instruction button 2012 to perform a search.

  FIG. 3 is a diagram illustrating an example of a search result in the search client. The search result is displayed by the search result display unit 202, and an outline of the search result is displayed by the summary display unit 203. The search result display unit 202 also serves as a document group specifying unit. By selecting an arbitrary number of articles using the document selection check box 2021 and clicking an associative search instruction button 2001, a document related to the selected article is searched. can do. The summary display means 203 also serves as a word group designation means. By selecting an arbitrary number of words using the word selection check boxes 2031 and 2032 and clicking the associative search instruction button 2001, a search from the summary words is performed. Can do.

  The associative search server 30 has a search request analysis unit 301 that analyzes a search request sent from the search client 20 and a search function that distributes the search request sent from the search client 20 to the search servers 40, 50, and 60. The request issuing unit 302 includes an approximate word requesting unit 303 that requests the search servers 40, 50, and 60 for an approximate word for a document group.

  The search request analysis unit 301 analyzes the search request sent from the search client 20, identifies the words included therein, and creates a search key. The search request analysis unit 301 may be arbitrary, but is configured to include at least a morpheme analysis that divides a sentence into words for a Japanese sentence, and a stemming process that performs original form reduction and part of speech addition for an English sentence. The

  The search request sent to the search request issuing unit 302 includes (1) a word set created by the search request analyzing unit 301 and (2) a search result display unit (document group specifying unit) included in the search client 20. A set of document IDs received, or (3) a set of words sent from the summary display means (word group designation means) 203 provided in the search client 20. When the search requests are (1) and (3), the word set is sent to the search server as a search request. When the search request is (2), the approximate word requesting unit 303 requests the search server for an outline of a document group corresponding to the set of document IDs, and the search server uses the sent approximate word set as a search request. Send to. The search server to which the search request issuing unit 302 sends the search request differs depending on the contents of the index held by each search server, and the operation will be described using an example described later.

  In the conventional associative search system, only one indexing is performed on one document database. An object of the present invention is to improve user convenience by indexing a single document database from a plurality of viewpoints. In order to realize this, the necessary requirements are (1) index creation from a plurality of viewpoints, (2) management of the same document included in a plurality of indexed document databases with a common identifier, It is. By managing the same document with a common identifier, it is possible to maintain the sameness between the indexes of the document set obtained as a search result, so that approximate words can be created for the same document set from each viewpoint. .

  4, 5, and 6 are examples of indexes when indexing from a plurality of viewpoints is performed on one document database base.

  FIG. 4 is an example in which the index of the document with the document ID 12345 is performed by a general term, a protein name, and an interaction between proteins. The number before each word in the index string is the frequency with which that word appears in the document. FIG. 5 is an example in which the index of the document whose document ID is 12345 is performed by the protein name. FIG. 6 shows an example in which the index of the document whose document ID is 12345 is performed by the protein-protein interaction. In order to satisfy the requirement (2), a common document ID “12345” is used in each indexing. The method of creating the index from each viewpoint may be arbitrary, but in reality, it is convenient to create so that one index includes a plurality of other indexes. In the above example, the index of FIG. 4 includes the indexes of FIGS. By doing so, all the search requests sent to the search request issuing unit 302 described above may be sent to the search server 40. The search servers 50 and 60 are used only when an outline is created for a search result.

  FIG. 3 is a diagram illustrating an example in which an associative search is performed using the indexes of FIGS. 4, 5, and 6. The article title is displayed as a search result. In addition, as an outline of the search results, protein names and protein-protein interactions included in these articles are displayed.

  Hereinafter, the flow of processing will be described with reference to the sequence diagrams of FIGS. For the sake of explanation, it is assumed that the indexes 404, 504, and 604 of the document databases 403, 503, and 603 provided in the search servers 40, 50, and 60 are created as shown in FIGS. 4, 5, and 6, respectively. When such indexing is performed, the operation of the search request issuing unit 302 is as follows. In response to the search request input by the user, the search request issuing unit 302 issues a search request to the search server 40. When creating an approximate word for the search result obtained from the search server 40, the approximate word requesting unit 303 issues a request for generating an approximate word to the search servers 50 and 60. When the user designates a document group and executes a re-search from the document group, a search request is issued to the search server 40. When the user designates a word group and executes a re-search from the word group, a search request is issued to the search server 40. In this way, all searches are performed by the search server 40. The search servers 50 and 60 are used only when creating approximate words of search results. Even when both the words “protein name” and “protein interaction” are designated, the search server 40 includes the indexes of the search servers 50 and 60, and thus operates without any problem.

  Next, the flow of processing will be described with reference to the sequence diagram of FIG. The user uses the search request input means 201 of the search client 20 to input a search request. The input search request is transmitted to the associative search server (T11). The search request analysis unit 301 of the associative search server 30 analyzes the search request and creates a search request for transmission to the search server. The search request issuing means 302 transmits a search request to the search server 40 (T12). The search means 402 of the search server 40 searches the document database 403 using the index 404 and transmits the result to the associative search server 30 (T13). The summary word requesting unit 303 of the associative search server 30 transmits a summary creation request for creating a summary of the obtained search results to the search server 50 and the search server 60 (T14, T16). The approximate word creation means 501 and 601 of the search server 50 and the search server 60 create approximate words using the indexes 504 and 604, respectively. In this example, the approximate word creation means 501 creates an approximate word composed of protein names, and the approximate word creation means 601 creates an approximate word composed of protein-protein interactions. The summary word created by each summary word creation means is transmitted to the associative search server 30 (T15, T17). Finally, the search result and the approximate word are transmitted from the associative search server 30 to the search client 20 (T18) and presented to the user by the search result display means 202 and the summary display means 203 of the search client 20.

  Next, description will be made with reference to the sequence diagram of FIG. This sequence diagram shows the flow of processing when a search is performed again from a document obtained as a search result or its approximate word.

  First, a case where re-searching is performed from a document obtained as a search result will be described. The user uses the document group specifying means 202 of the search client 20 to select a document that is a key for re-search. The identifier of the selected document is transmitted to the associative search server 30 (T21). The summary word requesting unit 303 of the associative search server 30 transmits a summary creation request for creating a summary of the selected document to the search server 40 (T22). The approximate word creation unit 401 of the search server 40 creates an approximate word using the index 404. That is, as described above, a statistically important word is selected and the approximate word is created by the same method as in Japanese Patent Laid-Open No. 2000-155758. The created approximate word is transmitted to the associative search server 30 (T23).

  When the user performs a search again only from the document, the obtained approximate word is transmitted to the search server 40 by the search request issuing unit 302 of the associative search server 30 (T25). The search means 402 of the search server 40 searches the document database 403 using the index 404 and transmits the result to the associative search server 30 (T26). The subsequent processing is the same as the processing after the rough word creating means in the sequence diagram of FIG.

  When the user performs a re-search from the approximate word, the user uses the word group designating unit 203 of the search client 20 to select a word as a key for the re-search. At this time, words from a plurality of viewpoints can be specified at the same time. The selected word or the identifier of the word is transmitted to the associative search server 30 (T24). The subsequent processing is the same as the processing after the search request issuing means in the sequence of FIG.

  By performing a re-search using an approximate word created from a certain viewpoint, the relationship between that viewpoint and another viewpoint can be grasped using the document database as an intermediary. For example, when a re-search is performed using a rough word composed of protein names, a document related to the selected protein name is obtained, and further, the protein name interaction related to the selected protein name is obtained. I can know. As a result, the search result can be analyzed in detail from various viewpoints.

  FIG. 9 shows an example in which a protein name and a disease name are used as indexes. By taking the same procedure as described above, the name of the disease associated with the protein name can be known from the name of the protein that the user is interested in. Conversely, the name of a protein associated with the name of the disease can be obtained from the name of the disease in which the user is interested.

Next, a modification of the present invention will be described with reference to FIG.
In the first embodiment, the viewpoint from which the outline of the search result is created is fixed in advance. However, it is also possible to prepare a plurality of search servers that hold indexes from a plurality of viewpoints in advance and select a viewpoint that the user wants to use. FIG. 10 is an example of an initial screen for the user to select a viewpoint.

  The viewpoint selection means 2013 presents three selectable viewpoints (index by gene “gene”, index by protein “protein”, protein interaction “protein interaction”) as viewpoints (view1, view2). . The user selects from which viewpoint he wants to get an overview. In the example of FIG. 10, the user has selected the index by the protein “protein” as view1 and the “protein interaction” as view2.

  Thereafter, the user inputs a search request in the search request input area 2011 and clicks a search instruction button 2012 to perform a search. The subsequent processing is the same as in the first embodiment.

Next, another modification of the present invention will be described with reference to FIG.
In the first embodiment, different servers hold indexes created from a plurality of viewpoints. That is, the index 404 of the search server 40 is held by the index of FIG. 4, the index 504 of the search server 50 is held by the index of FIG. 5, and the index 604 of the search server 60 is held by the index of FIG. However, a plurality of search servers are not necessarily required, and a single search server can hold a plurality of indexes.

  FIG. 11 is a configuration diagram when a plurality of indexes are held in one search server. In the document database 703 of the search server 70, indexes created from a plurality of viewpoints are held as indexes 704, 705, and 706. When holding a plurality of indexes in one search server, it is usual to hold each index independently. Each index can be in the form of a matrix with, for example, a document vertically and a word horizontally. In the elements of the matrix, appearance frequency information indicating how many times the word appears in the document is entered. In this case, since it is necessary to maintain the identity of the document on the vertical axis among a plurality of indexes (matrix), the same document is managed by the same identifier among the plurality of indexes.

  In the first embodiment, the search request issuing unit 302 included in the associative search server 30 controls which search server the search request is issued in accordance with the type of the search request. As shown in FIG. 11, when there is one search server, the search request issuing unit 302 may control which index of the search server 70 is used for searching according to the type of search request. . In the sequence diagrams of FIGS. 7 and 8, the same processing as in the first embodiment is performed by regarding all the search servers as the same search server.

1 is a schematic diagram showing a configuration example of a system for realizing the present invention. The figure which shows the example of the initial screen in a search client. The figure which shows the example of the search result in a search client. The figure which shows the example of indexing. The figure which shows the example of indexing. The figure which shows the example of indexing. The sequence diagram which shows the data between a search client, an associative search server, and a search server, and the flow of a process. The sequence diagram which shows the data between a search client, an associative search server, and a search server, and the flow of a process. The figure which shows the example of a display of the search result in a search client. The figure which shows the example of the initial screen in a search client. Schematic which shows the other structural example of the system for implement | achieving this invention.

Explanation of symbols

10: Communication network 20: Search client 2001: Associative search instruction button 201: Search request input means 2011: Search request input area 2012: Search instruction button 2013: Viewpoint selection means 202: Search result display means (document group specification means)
2021: Document selection check box 203: Outline display means (word group designation means)
2031: Word selection check box 2032: Word selection check box 30: Associative search server 301: Search request analysis means 302: Search request issue means 303: Outline word request means 40: Search server 401: Outline creation means 402: Search means 403: Document database 404: Index 50: Search server 501: Outline creation means 502: Search means 503: Document database 504: Index 60: Search server 601: Outline creation means 602: Search means 603: Document database 604: Index 70: Search server 701 : Outline creation means 702: Search means 703: Document database 704: Index 705: Index 706: Index

Claims (8)

A search client comprising an input unit for inputting a search request, a search result display unit for displaying a searched document group, and a general display unit for displaying an outline of the searched document group;
A document database storing a plurality of indexed documents, a search unit that searches the document database for documents highly relevant to a received search request, and an outline using the index for a given document group A search server including a schematic creation unit for creating
A document search system comprising a plurality of different types of indexes as the index.   2. The document search system according to claim 1, comprising a plurality of search servers, each search server having a different type of index, and the same document is managed by the same identifier among the document databases of the plurality of search servers. Document retrieval system characterized by that.   2. The document search system according to claim 1, wherein one search server includes a plurality of different types of indexes, and the same document is managed by the same identifier among the plurality of indexes. system.   2. The document search system according to claim 1, wherein one of the plurality of indexes is an index obtained by integrating the remaining plurality of indexes.   2. The document search system according to claim 1, wherein the summary display unit of the search client includes a summary display unit for each index that displays different summaries corresponding to different indexes.   6. The document search system according to claim 5, wherein the search client includes means for selecting a schematic element displayed on the schematic display unit, and transmits the selected element as the search request. Document retrieval system. A document database storing multiple documents;
A plurality of types of indexes assigned from different viewpoints to documents in the document database;
A search unit that searches the document database for documents highly relevant to the received search request;
A summary creation unit for creating a plurality of types of summary using the index for a given document group,
The search server, wherein the same document is managed by the same identifier among the plurality of indexes. An input part for inputting a search request;
A search result display section for displaying a document group as a received search result;
An outline display unit for displaying an outline of the document group for each of a plurality of different viewpoints;
The search result display unit has a document selection unit for selecting a document to be a key for the next search from the displayed document group,
The schematic display unit includes a schematic selection unit for selecting an element that is a key for the next search from the displayed schematic elements,
A search client, comprising: a search request input to the input unit, a document selected by the document selection unit, or information on a general element selected by the general selection unit as a search request.

Images